On Customizing Prosody in Speech Synthesis: Names and Addresses as a Case in Point